Development of KEK/HPK n+-in-p Pixel Sensor Modules and Understanding Their Performance with TCAD Simulations Y. Unno (KEK) for ATLAS-Japan Silicon Collaboration and Hamamatsu Photonics K.K. TREDI2015, 2015/2/18, Y. Unno1

Hybrid Planar Pixel Sensor Module Frontend ASIC and Pixel sensor can be optimized – independently, without compromise... We need 3 ingredients in the sensor: – Radiation tolerant pixel sensor (pursuing Planar process pixel sensor) – Bump-bonding (SnAg solder bump, e.g.) Also, thin sensor (≤150 µm) – thin ASIC (≤150 µm) – High voltage protection at edges against HV ~1000 V TREDI2015, 2015/2/18, Y. Unno2 Lightning?

Content Radiation-tolerant Planar Pixel Sensor – Identification of inefficient regions in irradiated sensors – Optimization of pixel structure(s) Understanding underlying physics of the inefficiency – with technology CAD simulation Not covered... – Bump-bonding – Edge protection TREDI2015, 2015/2/18, Y. Unno3

KEK/HPK n-in-p Pixel Sensors n-in-p 6” #2 wafer layout (“Old” pixel structures) FE-I4 2-chip pixels FE-I4 1-chip pixels FE-I3 1-chip pixels FE-I3 4-chip pixels n-in-p 6” #4 New wafer layout (“New” pixel structures) TREDI2015, 2015/2/18, Y. Unno4

“Old” Pixel Structures Severe efficiency loss at the boundary of pixels, under bias rail Subtle efficiency loss due to the routing of bias resistor TREDI2015, 2015/2/18, Y. Unno5 Irradiation: n 1×10 16 neq/cm 2 at Ljubljana V Bias railNo bias rail K. Motohashi et al. HSTD9 (DOI: /j.nima ) PolySi routing Irrad. sensors

Old Pixel Structures (Wafer #2) Bias rail → at the boundary of pixels Bias resistor (PolySilicon) → encircled outside the pixel implant Bias resistor and Bias rail are connected to the pixel electrode in DC, – thus, both are at “ground potential” TREDI2015, 2015/2/18, Y. Unno6 (#2 Wafer Layout) Pixel implant/electrode p-stop Bias rail Bias resistor

Optimization of Pixel Structures Bias rail → Removing from the boundary to “inside” the pixel electrode. – Removing “ground potential” at the boundary. PolySilicon bias resistor → routing inside the pixel. – Removing another “ground potential” outside the pixel. TREDI2015, 2015/2/18, Y. Unno7 (#2 Wafer Layout) (#4 Wafer Layout)

Bias Rail & Resistor Routing in Wafer#4 Bias rails away from the boundary: Large-, Small-, Zig-zag-offset – Bias rail material (Al, PolySi) Bias rail at the boundary but with “wide” p-stop Biasing structure: PolySi, Punch-Thru (PT) resistor, No biasing TREDI2015, 2015/2/18, Y. Unno8 PolySiPT Type 19 (No Bias) Type13 (Wide p-stop) Bias rail Type17 (Small-offset) Type18 (Large offset) Type2, 10 (Large-offset) Type4, 12 (Small-offset)

Evaluation of New Pixel Structures Irradiation at CYRIC – 70 MeV protons, Tohoku Univ., Japan – 3 to 5 x neq/cm 2 Latest setup – Irradiation box with 15 “push-pull” slots – “Liquid” Nitrogen cooling evaporated-in-supply-line TREDI2015, 2015/2/18, Y. Unno9 Samples in the irradiation box at CYRIC

Results with Testbeams TREDI2015, 2015/2/18, Y. Unno10 New design Old design Irrad. KEK46 (Type10) Comparison of “Area” – Width of the dip is due to pointing resolution, – Eliminating the effect of resolution. Left-Right imbalance is very much improved. – Bias rail effect is nearly eliminated. 1200V 100V 400V CERN testbeam DESY testbeam D. Yamaguchi

Comparison of Structures TREDI2015, 2015/2/18, Y. Unno11 Type19 (320µm) Type10 (150µm) Type13 (wide p-stop) (320µm) Scaled to 150 µm, 5x10 15 irrad. PT/p-spray Type10 (320µm) “Old” design “Old” design loses 2-3% eff. under the bias rail; 97-98% eff. overall. “New” design (Type10 (large offset)) is nearly as good as “no bias”, almost 0% loss >400 V. Type13 (wide p-stop) has been improved. D. Yamaguchi

TCAD Geometry Non- irrad Irrad Si thickness (µm) 150 Fluence (neq/cm 2 ) Null N eff (p- type)(cm -3 ) V dep (V app ) (V)44 (100)430 (430) Interface charge Q f (cm - 2 ) Pixel electrodeBias rail (Not to scale) TREDI2015, 2015/2/18, Y. Unno12 V dep = depletion voltage V app = applied bias voltage

Electron Layer Attracted to the interface charge, creation of an inversion layer of “electrons” is assumed. The layer has been simulated in TCAD. Irrad., with bias rail Electron layer disappears near the p-stop (p- stop edge at -2 µm). TREDI2015, 2015/2/18, Y. Unno13

Effect of Potential of Bias Rail Electric field (Potential) between pixels – near the surface (1 µm below the surface of Si in TCAD) Existence or non-existence of bias rail (“ground potential”) – has not affected the electric field potential very much. – Relative potential “depth” at the boundary is shallower in “Irrad.”: ~15% (=-15/-100, Non irrad.), ~9% (=-40/-430, Irrad.), but – Absolute potential is larger in “Irrad.”: -15 V (Non-irrad.), -40 V (Irrad.) Non-irrad. Irrad. With bias rail (black) No bias rail (Green) TREDI2015, 2015/2/18, Y. Unno14

Induced Charge – Ramo’s theorem A mobile charge in the presence of any number of grounded electrodes, the induced charge Q A at an electrode A is – where q is the charge in a position, V qA the “weighting potential” of the electrode A at the position of q. If a charge q moves along any path from position 1 to position 2 (after infinite time), In a finite time and with a readout circuitry, instantaneous induced current, i A, shall be integrated (with a proper shaping time) along the moving direction. (From V. Radeka) TREDI2015, 2015/2/18, Y. Unno15

Induced Charge – Ramo’s theorem We have to think two different fields: the “electric field” E x (and in turn the “electric field potential” V x ) and the “weighting potential” V qA. Although the final answer shall be obtained after integrating the current, we can have insight qualitatively from the relevant potentials, V x, E x, V qA TREDI2015, 2015/2/18, Y. Unno16

Non-irrad, With Bias Rail TREDI2015, 2015/2/18, Y. Unno17 V qA (Pixel) V qA (Bias rail) ExEx VxVx

Irrad, With Bias Rail VxVx ExEx V qA (Pixel) V qA (Bias rail) TREDI2015, 2015/2/18, Y. Unno18

Irrad., No Bias Rail VxVx ExEx V qA (Pixel) TREDI2015, 2015/2/18, Y. Unno19 Electric flux line Charges move along the electric flux lines.

Irrad., Wide P-stop VxVx ExEx V qA (Pixel) V qA (Bias rail) TREDI2015, 2015/2/18, Y. Unno20

V qA (Pixel) V qA (Bias rail) What is going on with bias rail? V qA (Pixel) V qA (Bias rail) V qA (Pixel) V qA (Bias rail) Non-irrad. Irrad. Irrad., No bias rail Irrad., Wide p-stop “Weighting potential” – of the “bias rail” has larger area of non-uniformity under the bias rail in “irrad.” than “non-irrad.” condition. Why? TREDI2015, 2015/2/18, Y. Unno21

Non-irrad. (Q f = ) Non-irrad. (Q f = ) Irrad. (Q f = ) Irrad. (Q f = ) Who is the Suspect? Strong electric field in the “irrad.” device has enhanced the non-uniform area of the weighting potential of the bias rail. Interface charge increase is acting to reduce the non-uniform area. Weak spot in the “non-irrad.” device deflects electric flux lines and the weighting pot., like a “shield”. TREDI2015, 2015/2/18, Y. Unno22 Irrad. V a;pp =-430 V Non-irrad. V app =-100 V V pA of Bias rail ExEx

Summary Novel design of the pixel structure has improved the efficiency loss due to the bias rail and bias resistor routing. – More structures need to be evaluated with testbeam to complete the variation of the design. Underlying physics has been understood with TCAD simulation (at least qualitatively). – The less-charge collection under the bias rail of the irrad. device seems to be caused by the fact that the bias rail is acting as (charge collecting) electrode. – Strong electric field has enhanced the charge to the bias rail. – Interface charge increase is helping to “reduce” the charge to the bias rail, (contrary to naive expectation). – In “non-irrad.” device, the bias rail as an electrode seems to be “shielded” with low electric field region under the bias rail. TREDI2015, 2015/2/18, Y. Unno23

Contributors ATLAS-Japan Silicon Group – KEK, Tokyo Inst. Tech., Osaka Uni., Kyoto Uni. Edu., Uni. Tsukuba, Waseda Uni. Hamamatsu Photonics K.K. PPS collaboration – AS CR, Prague, LAL Orsay, LPNHE / Paris VI, Uni. Bonn, HU Berlin, DESY, TU Dortmund, Uni. Goettingen, MPP and HLL Munich, Uni. Udine- INFN, KEK, Tokyo Inst. Tech., IFAE-CNM, Uni. Geneve, Uni. Liverpool, UC Berkeley, UNM- Albuquerque, UC Santa Cruz TREDI2015, 2015/2/18, Y. Unno24

Backup Slides TREDI2015, 2015/2/18, Y. Unno25

ATLAS Tracker Layouts Current inner tracker – Pixels: 5-12 cm Si area: 2.7 m 2 IBL(2015): 3.3 cm – Strips: (B)/28-56 (EC) cm Si area: 62 m 2 – Transition Radiation Tracker (TRT): cm Occupancy is acceptable for <3x10 34 cm -2 s -1 Phase-II at HL-LHC: 5x10 34 cm -2 s - 1 Phase-II upgrade (LOI) – Pixels: 4-25 cm Si area: 8.2 m 2 – Strips: (B) cm Si area: 122 (B)+71(EC)=193 m 2 Major changes from LHC – All silicon tracker – Large increase of Si area both in Pixels and Strips ~ 3 × LHC ATLAS TREDI2015, 2015/2/18, Y. Unno26

ATLAS detector to design for – Instantaneous lum.: 7x10 34 cm -2 s -1 – Integrated lum.: 6000 fb -1 (including safety factor 2 in dose rate) – Pileup: 200 events/crossing Particle fluences in HL-LHC PIXELs (HL-LHC) – Inner: r=3.7 cm ~2.2x10 16 – Medium: r = 7.5 cm, ~6x10 15 – Med/Out: r=15.5 cm ~2x10 15 – Outer: r = 31 cm (?) ~1x10 15 – Charged:Neutrons ≥ 1 STRIPs (HL-LHC) – Replacing Strip and TRT – Short strip: r = 30 cm, e.g. ~1x10 15 – Long strips: r = 60 cm, ~5×10 14 – Neutrons:Charged ≥ 1 IBL (LHC) – Insertable B-layer pixel – r = 3.3 cm Flunece ~3x10 15 neq/cm 2 at Int.L~300 fb -1 TREDI2015, 2015/2/18, Y. Unno27 Short strips Long strips

Bias Rail Routing Bias rail to offset from midway: Large, Small Bias Type: PolySi, Punch-Thru (PT) Number of bias rail: Single, Double, None Bias rail material (Al, PolySi) TREDI2015, 2015/2/18, Y. Unno28 PolySi PT No Bias Type (1, 9) Type (2, 10) Type (3, 11) Type (4, 12) Type 17 Type 18 Type 19

Narrow pitch 25 µm pixels TREDI2015, 2015/2/18, Y. Unno29 Type 21 Type 22 Type20 PT No offset Type 23 PT Offset End1 End2 Type 25 Staggered PolySi Type 28 No Bias Bump pads at the midway of the pixels Bias rail offset: No, Offset Bias: PolySi, PT, No bias Bias rail material (Al, PolySi)

Pixel Structures in #4 Wafer For your reference, TREDI2015, 2015/2/18, Y. Unno30

Edge Width for holding 1000 V S. Mitsui et al., NIMA699(2013)36-40 TREDI2015, 2015/2/18, Y. Unno µm Edge width varied CYRIC irradiations 70 MeV proton

Why less induced charge at the bias rail in “Non-irrad.” condition? Why are the induced charges at the bias rail in “non-irrad.” and “irrad.” condition different? – Isn’t this against natural intuition? There is weak electric field region in the silicon and the p-stop under the bias rail. – larger in “non-irrad.”, smaller in “irrad”. – Electric flux line has been deflected with the weak field region. – The weak electric field seems to “shield” the bias rail as electrode. TREDI2015, 2015/2/18, Y. Unno32 ExEx Non-irrad. (V app =-100 V) ExEx Irrad. (V app =-430 V)

Non-irrad. (Q f = ) Non-irrad. (Q f = ) Irrad. (Q f = ) Who is the Suspect? Primary: Strong electric field – creates non-uniform weighting potential of the bias rail. Secondary: Interface charge – alters the electric field and is reducing the weighting potential TREDI2015, 2015/2/18, Y. Unno33 V app =-100 V V a;pp =-430 V V app =-100 V V pA of Bias rail